Problem with arrays structures sorting.

I am trying to write a program that reads elements and their relative abundances from a file, then to sort them alphabetically or in order of abundance. however i have been having problems with the part involving structures and arrays to read the data from the file and get it into a format that can be sorted in the desired manner.

My code so far is

Code:

---

#include <stdio.h>
#include <stdlib.h>

int main()
{
char name[128];
FILE *f;
char s[1000];
printf("\nPlease input a file name including extension\n");
scanf("%s",name);
f=fopen(name,"r");
if (!f) return 1;
  while (fgets (s,1000,f)!=NULL)
  printf("%s",s);
  fclose(f);
/*sorting*/
 
        printf("how would you like the data sorted?\n press 1 if you would like the data sorted alphabetically by name of element, or press 2 to sort data by ascending percentage abundance\n ");
        scanf("%d", &y);

/*sorting functions*/

                if(y == 1) /*alphabet*/
        {
         


        }

        else if (y == 2) /*number*/
        {


        }
 
 

  return 0;

}

---

a sample input is:

Oxygen 15
Carbon 60
Hydrogen 2
Nitrogen 20
Boron 3

if anyone has any pointers or help to offer, i would be very grateful.
pitifulworm.

All element names are one word, and numbers are numbers, so you should be able to use the formatted read functions (like *scanf) to read from your file into ... whatever you want to put them in. Presumably you would make an array of structs, where each struct would contain some chars for the name and an int for the int.

Structures would be a good way to go, another good, and simpler way to go, would be to use parallel arrays:

element[N][n] atoms[N]

Where N would correspond in the element array, with the same N in the atom array. The small n would be a length that would exceed any word in your list - 20 should surely do.
(what element has the longest name? californium?

Data would look like:

Code:

---

Hydrogen  1 
Helium    2
Boron    5

---

etc.

after it was sorted by number of atoms. Edit: we're sorting by abundance, not number of atoms.

In the sorter, when a swap is needed, you need to swap both the name of the element, in the element[] *and* the atom[], to keep the proper alignment of name with number, in place.

it's easier to sort an array of structures than parallel arrays, so that bit of complexity is worth it. you can also use qsort instead of making up a sort algorithm.

Parallel arrays are taught just before structures, in nearly all intro CS classes.

pitifulworm mentioned structures, so what's the problem?

Quote:

---

however i have been having problems with the part involving structures

---

The OP doesn't know how to work with structures, yet. This assignment "smells" like a parallel array problem, to me.

I have no problem if the OP wants to use a structure, at all. Always more than one way to get the job done.

oh, i don't want to get in the way of your sense of smell then. :rolleyes:

Pitifulworm, you need to chime in here. If the assignment requires the name of the element, the number of atoms, *and* anything else (like it's abundance in a file), then you need a structure to hold that info.

If your assignment just requires the name of the element and the number of atoms, then a parallel array is fine, OR a structure, either one.

Both can be sorted, loaded, etc., with very little difference in terms of work involved.

So, which way do you prefer?

the sample data is the name of the element and its % abundance, so i need these to stay together so that it can sort in order of abundance or alphabetically. i am pretty new to c programming so haven't done structures and arrays much, so which would be better; a parrallel array (of which i have not done at all ever) or i could just do an array of structures?

They're really about the same. (despite our debate on it in the thread). IMO parallel arrays is a better choice if you have two things to keep together. For three or more things, a struct is the choice that makes sense.

Quote:

---

Originally Posted by pitifulworm

i am pretty new to c programming so haven't done structures and arrays much, so which would be better; a parrallel array (of which i have not done at all ever) or i could just do an array of structures?

---

try them both. the worst that could happen is you learn another way to solve the problem.

Quote:

---

Originally Posted by Adak

IMO parallel arrays is a better choice if you have two things to keep together.

---

except that you have to manage two entities instead of one, and if you're sorting you can't use qsort anymore. imo when ppl can't use qsort they resort to something crappy and easy to write like bubble sort. parallel arrays pretty much always make the code worse in one way or another.

You can use qsort with parallel arrays.

Just swap the corresponding second array's element whenever you swap an element from the first array (the key array). That's the whole extent of "managing the entity". LOL

Quote:

---

Originally Posted by Adak

You can use qsort with parallel arrays.

Just swap the corresponding second array's element whenever you swap an element from the first array (the key array).

---

show me the code, please. btw, i'm talking about the qsort function in stdlib.h, not your own ad hoc quicksort function.

Quote:

---

Originally Posted by Adak

Just swap the corresponding second array's element whenever you swap an element from the first array (the key array).

---

hmm... but you only provide the comparator and qsort() performs the swapping. I do not see how it can be done without changing the implementation for qsort() to support the parallel arrays since qsort() does not use a callback function to perform swapping.

I believe C.A.R. Hoare would be very upset if we had to use just one implementation of his beautiful sorting algorithm. :)

Did you jump to a conclusion there Meldreth? You know what happens when you jump to conclusions like that. :p :p

Echo my left foot. :)

Quote:

---

Originally Posted by Adak

I believe C.A.R. Hoare would be very upset if we had to use just one implementation of his beautiful sorting algorithm.

---

without lots of proof, i trust the library more than i trust an ad hoc implementation by an egotist who thinks he can compete.

Quote:

---

Did you jump to a conclusion there Meldreth?

---

i could say you jumped to the conclusion that when i wrote qsort it meant the quicksort algorithm and not c's qsort function. i think what really happened is you intentionally misread my words so you could make a petty jab at me. and you proved it with your next statement. weak.

Quote:

---

Originally Posted by Adak

Echo my left foot.

---

bitter much? :rolleyes:

Quote:

---

Originally Posted by Adak

I believe C.A.R. Hoare would be very upset if we had to use just one implementation of his beautiful sorting algorithm.

---

Hoare's sorting algorithm is named "quick sort", not "qsort".

Quote:

---

Originally Posted by Meldreth

without lots of proof, i trust the library more than i trust an ad hoc implementation by an egotist who thinks he can compete.

i could say you jumped to the conclusion that when i wrote qsort it meant the quicksort algorithm and not c's qsort function. i think what really happened is you intentionally misread my words so you could make a petty jab at me. and you proved it with your next statement. weak.

bitter much? :rolleyes:

---

No no, just having fun. I thought you were, as well.

Qsort is quite good, but it's not the very best implementation of Quicksort. The newer C++ version is one such better version.

To be honest, I was thinking of my Quicksort code, but of course, it's not *my* code. It's from a book. I used to tweak it a bit for special sorting jobs. Don't do that anymore, though. There's no need to tweak it, because the faster hardware has made all but the most extremely demanding sorting job, well within what I need.

I'm just a hobby coder, so I'm guessing you're a better programmer than I am. I'm not competing with you, or the standard library version of Quicksort.

Quote:

---

Originally Posted by Adak

Qsort is quite good, but it's not the very best implementation of Quicksort. The newer C++ version is one such better version.

---

As far as I can tell, quicksort is not mandated for the implementation of qsort and std::sort, though typically a variant of quicksort is used.

Quote:

---

Originally Posted by Adak

No no, just having fun. I thought you were, as well.

---

of course i'm havng fun. you're my new "Special Friend". ;)

Quote:

---

Originally Posted by Adak

Qsort is quite good, but it's not the very best implementation of Quicksort.

---

so you've checked the source code for every compiler in existence and determined that none of them uses the very best implementation of quicksort? who's measure of "very best" are you using? c++'s? c++ usually uses introsort, and introsort isn't really quicksort. it's a hybrid of quicksort, heapsort, and insertionsort designed to overcome quicksort's flaws. there's nothing stopping a compiler from using introsort for qsort too. or bubblesort. at least c++'s sort has a performance guarantee.

Quote:

---

Originally Posted by Adak

I'm just a hobby coder, so I'm guessing you're a better programmer than I am.

---

damn skippy. ;) who knows or cares who's better, but my ego dictates that i have to assume it's always me.

I read a book on sorting, and it showed how quicksort could be optimized, if you were careful.

I don't have that ego driven dictate. It would make no sense to think I'm somehow "better", because I clung to some vaporous assumption. I like to be more objective than that.

I recall an episode of the Animaniac's where they were being canceled, and they did this long and very hilarious song about their studio exec. I was in stitches. Never saw a funnier cartoon. Their 50 states and capitals song was quite clever, as well.

Meldreth's "animaniacs" observation fits my theory that quicksort would really be better known as quacksort -- but I said that somewhere else already.

"introsort" must really be the beginning of the end.

Quote:

---

Originally Posted by MK27

Meldreth's "animaniacs" observation fits my theory that quicksort would really be better known as quacksort -- but I said that somewhere else already.

---

I really don't see why any correct sorting algorithm should be known as "quacksort", except maybe bogosort, but even bogosort is correct.

Could give me an example to use as i am strugglling to start off, i know this is asking an annoying question but i have dedicated a lot of time to this and it is getting very frustrating.
thanks

Code:

---

#include <stdio.h>
#include <stdlib.h>
#include <string.h>

struct element
{
    char name[128];
    int freq;
};

int cmp_name(const void *a, const void *b);
int cmp_freq(const void *a, const void *b);
void list(struct element *a, int sz);

int main()
{
    FILE *fp = fopen("test.txt", "r");
    int (*cmp_pf)(const void*, const void*);
    struct element elm[200];
    int n, opt;
    for (n = 0; n < 200; n++)
        if (fscanf(fp, "%s%d", elm[n].name, &elm[n].freq) != 2)
            break;
    fclose(fp);
    printf(
        "how would you like the data sorted?\n"
        "1: by name\n"
        "2: by abundance\n> ");
    scanf("%d", &opt);
    cmp_pf = (opt == 1) ? &cmp_name : &cmp_freq;
    qsort(elm, n, sizeof(struct element), cmp_pf);
    list(elm, n);
    return EXIT_SUCCESS;
}

int cmp_name(const void *a, const void *b)
{
    return strcmp(((struct element*)a)->name, ((struct element*)b)->name);
}

int cmp_freq(const void *a, const void *b)
{
    return ((struct element*)a)->freq - ((struct element*)b)->freq;
}

void list(struct element *a, int sz)
{
    int i;
    for (i = 0; i < sz; i++) printf("%s %d\n", a[i].name, a[i].freq);
}

---

Thank you so much, that has helped me get my head around how to work my program.

I have made a draft of the program using your code as a basis, however i keep getting either an error when runnign from command promt saying " the instruction at 0x7c90100b referenced memory at 0x00000034. the memory could not be read" or if i run through my compiler i get a screen continually displaying pages of 1/4 over again. can;t seem to work out the problem, i'm probably overlooking somthing aren't i?
here is the code:

Code:

---

#include <stdio.h>
#include <stdlib.h>
#include <string.h>

struct element
{
    char name[128];
    int abund;
};

int cmp_name(const void *a, const void *b);
int cmp_abund(const void *a, const void *b);
void list(struct element *a, int sz);

int main()
{
   
    char name[128];
    FILE *fp;
    char s[1000];
    {
    printf("\nPlease input a file name including extension\n");
    gets(name);
    fp=fopen(name,"r");
    while (s!=NULL)
    printf("%s",s);
    while (fgets (s,1000,fp)!=NULL);
    fclose(fp);
    }
 
    fp=fopen(name,"r");
    int (*cmp_pf)(const void*, const void*);
    struct element elm[1000];
    int n, opt;
    for (n = 0; n < 1000; n++)
        if (fscanf(fp, "%s%d", elm[n].name, &elm[n].abund) != 2)
            break;
    printf(
        "how would you like the data sorted?\n"
        "1: by name\n"
        "2: by abundance\n> ");
    scanf("%d", &opt);
    cmp_pf = (opt == 1) ? &cmp_name : &cmp_abund;
    qsort(elm, n, sizeof(struct element), cmp_pf);
    list(elm, n);
    return EXIT_SUCCESS;
}

int cmp_name(const void *a, const void *b)
{
    return strcmp(((struct element*)a)->name, ((struct element*)b)->name);
}

int cmp_abund(const void *a, const void *b)
{
    return ((struct element*)a)->abund - ((struct element*)b)->abund;
}

void list(struct element *a, int sz)
{
    int i;
    for (i = 0; i < sz; i++)
    printf("%s %d\n", a[i].name, a[i].abund);
}

---

An error with such a low memory address means you've accessed something relative to a NULL pointer.

"%s%d" does not sound like what you want for your fscanf. Don't you have a seperator such as a comma or space between those fields? I haven't run the program but I'd guess the bug is on that line.
Try printing n after the for loop.

Probably a good idea to read this (Meldreth too):
http://www.akalin.cx/2006/06/23/on-t...ison-function/

nope, same problem occurs

replace the return in cmp_name function with this:

Code:

---


in this function:

int cmp_name(const void *a, const void *b)
{
    return(strcmp(a,b));
    //return strcmp(((struct element*)a)->name, ((struct element*)b)->name);
}

This is an elegant program, but clearly several steps above where you are with C. Nice as it is, I'd look
for code help in the future, that was more at your level, or just above it.

I don't think you learn much from turning in someone else's program "whole hog". This program looks
nothing like a program from a beginner.

---

and it works ok, in very limited testing.

Quote:

---

Originally Posted by pitifulworm

nope, same problem occurs

---

That's a completely meaningless response!
You haven't demonstrated that you've even understood the advice given, you haven't answered the question, and you didn't say what the value of n was when you printed it out after the loop.
I haven't got a clue what change you might have made.
How about posting a sample of the contents of your data file too!

You are definitely someone who needs to read "How to ask questions the smart way".

She previously posted this as her input:

Quote:

---

a sample input is:

Oxygen 15
Carbon 60
Hydrogen 2
Nitrogen 20
Boron 3

---

Making the one line change I noted above, causes the program she posted, to work, in very limited testing.

OP, a tactic I use when debugging is to comment out everything apart from main and return. Then slowly bring things back in to the code and see where the error lies.

Another tactic I use when it compiles and runs but doesn't get all the way through is to simply put a few printf lines in and see which ones it prints and which ones it doesn't print.

Good luck in your CSP

Quote:

---

Originally Posted by Adak

replace the return in cmp_name function with this:

Code:

---


in this function:

int cmp_name(const void *a, const void *b)
{
    return(strcmp(a,b));
    //return strcmp(((struct element*)a)->name, ((struct element*)b)->name);
}

This is an elegant program, but clearly several steps above where you are with C. Nice as it is, I'd look
for code help in the future, that was more at your level, or just above it.

I don't think you learn much from turning in someone else's program "whole hog". This program looks
nothing like a program from a beginner.

---

and it works ok, in very limited testing.

---

you're passing struct element pointers to strcmp. that's a really bad idea because now the comparison routine depends on the order of members in the struct. say after testing you find out that you can save some memory by putting the int member first.

Code:

---

struct element
{
    int abund;
    char name[128];
};

---

now the comparison will be messed up and you have to chase down a bug that's completely unrelated to the change you made.

I'm with meldreth here. There are many ways that this could go wrong - if you move it to C++ it would fall apart if you introduce a virtual function [assuming with no other changes].

--
Mats

I'm sure you could "break" any program a thousand different ways - but the version of Meldreth's program that the OP put up, wouldn't run in Turbo C, either.

With that one line change, it now runs OK. If hand waving and what if's would fix it, I'd be all for them.

it's not about breaking a program a thousand different ways. it's about making those breaks easier to find. your way is hard to find if they happen. i don't use turbo crap anymore so i can't give you any hand waving to make the code work, but i *can* tell you from experience that your bug will be a royal pain in the ass to troubleshoot. you need to find something else that compiles on your compiler or get a compiler that understands c better.

I didn't run your program as you posted it, but I thought it looked good.

When the OP came back and said it wouldn't run, then I tried it out, and found a fix for the program the OP posted up. What the OP posted up was perhaps not quite the same as your program.

I have two version of Visual C/C++ (6.0 and 9), and an Intel compiler as well. They all have their uses.

Whether it's a program, or a car, it's more important to keep it running, than to keep it easy to service. That has to be a secondary consideration.

Quote:

---

Originally Posted by Adak

Whether it's a program, or a car, it's more important to keep it running, than to keep it easy to service.

---

and just like with a car, easy to service means lower maintenance costs to keep it running.

Ok, so first of all, the code posted [before the first one before two iMalc posts back] is broken because it printf's forever the uninitialized variable s. Removing that, it fails to sort the values in "abundance order" with an access to 0x34. It sorts correctly in alphabetical order - at least with the given input above.

Trying it again, it doesn't appear to fail, so I'm not sure what is going on there....

--
Mats

I think i have got it working now, thanks to all for you help. now i've got to puts a whole lot of ifs in if the user enters crap...man.

Quote:

---

Originally Posted by Adak

She previously posted this as her input:

---

That'll teach me to make a last minute edit.